G-Code Error?

The M111/M112 are on/off vacuum clamping, and M113,M115,M123,M125 activate/deactivate control of the spindles on the W and Z heads. ( I commented them out on my simulation runs. For example if I were running a file using the Z spindle, M113 would allow M3 and M5 to turn that spindle on and off. M123 would connect the spindle on the W head to M3/M5. M115/M125 dis-connect the Z/W spindles from the M3/M5. If I were running a file with the W spindle only, I would issue M115 and M123 (then M3/M5 would control only the W spindle), running both spindles would use M113 and M123 (then M3/M5 would control both spindles). I don't think they are contributing to the problem.

I agree the custom M-codes are not related to this bug. I can reproduce the problem here without them. I'm looking into it more.

I just pushed a fix for this bug. It'll be in v2.7.0-39-g1c5461c from the buildbot, and in 2.7.1 when that gets released.

What was causing the problem?

It was a bug introduced between 2.6 and 2.7, as part of the trajectory planner work.

When doing arcs in the XY plane, the secondary axes (ABCUVW) could sometimes be converted from program coordinates to a simplified internal representation called "canonical coordinates" twice, resulting in the bogus W value that you reported.

Give the 2.7 branch a try and let me know if it fixes the problem for you.

Thanks

Looking forward to trying it.

Is it only available through git at the moment?

When should it be ready from buildbot? It may be a week or so until I might have the time to setup a git pull on one of the machines.

Packages with the fix are available from the buildbot now. Make sure you get v2.7.0-39 or newer.

Yes that seems to have fixed the problem for me.

Thank you for all your efforts. Even with the few bugs that are there, I think 2.7 (and it's new TP) are a vast improvement over 2.6 for us.